“Uses high-stakes keywords like 'bioweapons' and 'looming threat' to frame AI risks as a serious, imminent danger.”